AI assisted feedback system for transmit parameter optimization in Cardiac Ultrasound

Ultrasound image acquisition is often not repeatable with dependency on the skill of the operator to both acquire and optimize ultrasound images. Beyond arriving at the right scan planes, the user has to optimize a series of acquisition parameters like the frequency, depth, gain, TGC etc to obtain the optimal image. While varying acquisition parameters leads to varying image quality and clinical acceptability, image quality in ultrasound is very subjective and thus the choice of optimal image is often limited to the preference of the user. In this paper, we present an AI assisted feedback system to optimize ultrasound acquisition parameters. We also present a generic methodology for `Ultrasound Image Quality' computation mimicking experts' way of reading u/s images. We demonstrate this system to optimize two transmit parameters, namely frequency and depth on 4 chamber cardiac Ultrasound view images. We show that in >80% of unseen cardiac acquisitions, the proposed system is able to automatically select the optimal parameters, maximizing image quality.